Identifying Diagnostic Indicators for Type 2 Diabetes Mellitus from Physical Examination Using Interpretable Machine Learning Approach

Overview

Journal Front Endocrinol (Lausanne)

Specialty Endocrinology

Date 2024 Apr 2

PMID 38562414

Authors

Xiang Lv

Jiesi Luo

Wei Huang

Hui Guo

Xue Bai

Pijun Yan

Zongzhe Jiang

Yonglin Zhang

Runyu Jing

Qi Chen

Menglong Li

Affiliations

Soon will be listed here.

Abstract

Background: Identification of patients at risk for type 2 diabetes mellitus (T2DM) can not only prevent complications and reduce suffering but also ease the health care burden. While routine physical examination can provide useful information for diagnosis, manual exploration of routine physical examination records is not feasible due to the high prevalence of T2DM.

Objectives: We aim to build interpretable machine learning models for T2DM diagnosis and uncover important diagnostic indicators from physical examination, including age- and sex-related indicators.

Methods: In this study, we present three weighted diversity density (WDD)-based algorithms for T2DM screening that use physical examination indicators, the algorithms are highly transparent and interpretable, two of which are missing value tolerant algorithms.

Patients: Regarding the dataset, we collected 43 physical examination indicator data from 11,071 cases of T2DM patients and 126,622 healthy controls at the Affiliated Hospital of Southwest Medical University. After data processing, we used a data matrix containing 16004 EHRs and 43 clinical indicators for modelling.

Results: The indicators were ranked according to their model weights, and the top 25% of indicators were found to be directly or indirectly related to T2DM. We further investigated the clinical characteristics of different age and sex groups, and found that the algorithms can detect relevant indicators specific to these groups. The algorithms performed well in T2DM screening, with the highest area under the receiver operating characteristic curve (AUC) reaching 0.9185.

Conclusion: This work utilized the interpretable WDD-based algorithms to construct T2DM diagnostic models based on physical examination indicators. By modeling data grouped by age and sex, we identified several predictive markers related to age and sex, uncovering characteristic differences among various groups of T2DM patients.

Citing Articles

Enhancing Type 2 Diabetes Treatment Decisions With Interpretable Machine Learning Models for Predicting Hemoglobin A1c Changes: Machine Learning Model Development.

Kurasawa H, Waki K, Seki T, Chiba A, Fujino A, Hayashi K JMIR AI. 2024; 3:e56700.

PMID: 39024008 PMC: 11294778. DOI: 10.2196/56700.

References

Zhao M, Wan J, Qin W, Huang X, Chen G, Zhao X . A machine learning-based diagnosis modelling of type 2 diabetes mellitus with environmental metal exposure. Comput Methods Programs Biomed. 2023; 235:107537. DOI: 10.1016/j.cmpb.2023.107537. View

Matheus A, Tannus L, Cobas R, Palma C, Negrato C, Gomes M . Impact of diabetes on cardiovascular disease: an update. Int J Hypertens. 2013; 2013:653789. PMC: 3603160. DOI: 10.1155/2013/653789. View

Rijzewijk L, Jonker J, van der Meer R, Lubberink M, de Jong H, Romijn J . Effects of hepatic triglyceride content on myocardial metabolism in type 2 diabetes. J Am Coll Cardiol. 2010; 56(3):225-33. DOI: 10.1016/j.jacc.2010.02.049. View

Saxena R, Sharma S, Gupta M, Sampada G . A Novel Approach for Feature Selection and Classification of Diabetes Mellitus: Machine Learning Methods. Comput Intell Neurosci. 2022; 2022:3820360. PMC: 9033325. DOI: 10.1155/2022/3820360. View

Saxena R, Voight B, Lyssenko V, Burtt N, de Bakker P, Chen H . Genome-wide association analysis identifies loci for type 2 diabetes and triglyceride levels. Science. 2007; 316(5829):1331-6. DOI: 10.1126/science.1142358. View

Howard R, Buddington B, Alfrey A . Urinary albumin, transferrin and iron excretion in diabetic patients. Kidney Int. 1991; 40(5):923-6. DOI: 10.1038/ki.1991.295. View

Wong T, Cheung C, Larsen M, Sharma S, Simo R . Diabetic retinopathy. Nat Rev Dis Primers. 2016; 2:16012. DOI: 10.1038/nrdp.2016.12. View

Rudin C . Stop Explaining Black Box Machine Learning Models for High Stakes Decisions and Use Interpretable Models Instead. Nat Mach Intell. 2022; 1(5):206-215. PMC: 9122117. DOI: 10.1038/s42256-019-0048-x. View

De Silva K, Jonsson D, Demmer R . A combined strategy of feature selection and machine learning to identify predictors of prediabetes. J Am Med Inform Assoc. 2020; 27(3):396-406. PMC: 7647289. DOI: 10.1093/jamia/ocz204. View

10.

Eisner B, Porten S, Bechis S, Stoller M . Diabetic kidney stone formers excrete more oxalate and have lower urine pH than nondiabetic stone formers. J Urol. 2010; 183(6):2244-8. DOI: 10.1016/j.juro.2010.02.007. View

11.

Arbet J, Brokamp C, Meinzen-Derr J, Trinkley K, Spratt H . Lessons and tips for designing a machine learning study using EHR data. J Clin Transl Sci. 2021; 5(1):e21. PMC: 8057454. DOI: 10.1017/cts.2020.513. View

12.

Bhat S, Selvam V, Ansari G, Ansari M, Rahman M . Prevalence and Early Prediction of Diabetes Using Machine Learning in North Kashmir: A Case Study of District Bandipora. Comput Intell Neurosci. 2022; 2022:2789760. PMC: 9553420. DOI: 10.1155/2022/2789760. View

13.

Zhang M, Wang J, Wang W, Yang G, Peng J . Predicting cell-type specific disease genes of diabetes with the biological network. Comput Biol Med. 2023; 169:107849. DOI: 10.1016/j.compbiomed.2023.107849. View

14.

Carson A, Muntner P, Kissela B, Kleindorfer D, Howard V, Meschia J . Association of prediabetes and diabetes with stroke symptoms: the REasons for Geographic and Racial Differences in Stroke (REGARDS) study. Diabetes Care. 2012; 35(9):1845-52. PMC: 3424995. DOI: 10.2337/dc11-2140. View

15.

Marathias K, Agroyannis B, Mavromoustakos T, Matsoukas J, Vlahakos D . Hematocrit-lowering effect following inactivation of renin-angiotensin system with angiotensin converting enzyme inhibitors and angiotensin receptor blockers. Curr Top Med Chem. 2004; 4(4):483-6. DOI: 10.2174/1568026043451311. View

16.

Maalouf N, Cameron M, Moe O, Sakhaee K . Metabolic basis for low urine pH in type 2 diabetes. Clin J Am Soc Nephrol. 2010; 5(7):1277-81. PMC: 2893060. DOI: 10.2215/CJN.08331109. View

17.

Patro K, Allam J, Sanapala U, Marpu C, Samee N, Alabdulhafith M . An effective correlation-based data modeling framework for automatic diabetes prediction using machine and deep learning techniques. BMC Bioinformatics. 2023; 24(1):372. PMC: 10544445. DOI: 10.1186/s12859-023-05488-6. View

18.

Tasin I, Nabil T, Islam S, Khan R . Diabetes prediction using machine learning and explainable AI techniques. Healthc Technol Lett. 2023; 10(1-2):1-10. PMC: 10107388. DOI: 10.1049/htl2.12039. View

19.

Ye X, Kong W, Zafar M, Chen L . Serum triglycerides as a risk factor for cardiovascular diseases in type 2 diabetes mellitus: a systematic review and meta-analysis of prospective studies. Cardiovasc Diabetol. 2019; 18(1):48. PMC: 6466658. DOI: 10.1186/s12933-019-0851-z. View

20.

Reusch J . Diabetes, microvascular complications, and cardiovascular complications: what is it about glucose?. J Clin Invest. 2003; 112(7):986-8. PMC: 198532. DOI: 10.1172/JCI19902. View